Enacted Extraversion as a Well-Being Enhancing Strategy in Everyday Life: Testing Across Three, Week-Long Interventions

Personality traits are consistent individual differences in patterns of thoughts, feelings, and behaviors that persist over time (McCrae & Costa, 2003). Broadly, extraversion describes the tendency to engage in behavior that is bold, assertive, energetic, talkative, and adventurous, and is conceptualized as a bipolar construct on a continuum with introversion, characterized by behavior that is reserved, quiet, modest, and passive (Fleeson et al., 2002; Goldberg, 1990). The exact make-up of extraversion (facets or narrower sub-traits) varies among taxonomies, but all include elements beyond mere sociability, such as high activity, excitement seeking, and sometimes even positive affect (e.g., DeYoung et al., 2007; Goldberg, 1990; K. Lee & Ashton, 2018; McCrae & Costa, 2003).

The (sometimes definitional) link between extraversion and positive affect strongly suggests an association with well-being; moreover, the correlation extends across a wide variety of indicators (Zelenski, Sobocko, et al., 2013). Beyond positive affect (Lucas & Fujita, 2000), positive associations have been observed between extraversion and the presence of meaning in life (Steger et al., 2006), subjective vitality (Ryan & Frederick, 1997), and satisfaction with life (DeNeve & Cooper, 1998; Steel et al., 2008). Additionally, extraversion has been positively associated with many aspects of psychological well-being, including self-acceptance, environmental mastery, purpose in life, and positive relations with others (Cooper et al., 1995; Schmutte & Ryff, 1997; Steel et al., 2008). For example, in a recent meta-analysis of 462 studies (n = 334,567) the mean correlations between various indicators of subjective well-being and psychological well-being with trait extraversion, as measured using the ‘big-five’ and HEXACO approaches, were r = .37 and r = .48, respectively (Anglim et al., 2020).

Despite consensus about the descriptive link between extraversion and well-being, disagreement remains as to what may explain this relationship. Some draw on extraverts’ tendencies to socialize, arguing that these behaviors, and the relationships they foster, produce happiness. Extraverts spend less time alone than introverts (Argyle & Lu, 1990; Leary et al., 2003), feel more socially connected to others (R. M. Lee et al., 2008), have a greater number of sexual partners throughout the lifetime (Heaven et al., 2000), and are perceived by their peers as being more competent and more likable (Ntalianis, 2010; Wortman & Wood, 2011). However, it is possible that extraverts’ tendencies to socialize are more a consequence of happiness than its cause, or at least that the causal link is bidirectional. That is, many see the core of extraversion as being a high baseline level of positive affect (Lucas & Baird, 2004), or as a general sensitivity to reward that then facilitates frequent positive affect (Lucas et al., 2000; Smillie et al., 2012; Zelenski & Larsen, 1999).

Socializing is one way to become happier (Sandstrom & Dunn, 2014), but positive affect also promotes a desire to socialize (Whelan & Zelenski, 2012). When socializing is dissociated from pleasantness in scenarios, extraverts prefer only the pleasant contexts more strongly than introverts (Lucas & Diener, 2001). In day-to-day life, the amount of socializing only partially explains the link between extraversion and happiness (Lucas et al., 2008; Srivastava et al., 2008) and both extraverts and introverts report higher levels of positive affect when engaging in social situations (Lucas et al., 2008). As such, it seems likely that there are extraverted paths to happiness that occur independent of socializing.

In contrast with personality traits, personality states are defined as momentary behaviors that can also characterize the same thoughts, feelings, and behaviors as the overall personality trait, but on a temporary basis (e.g., over the course of minutes or hours; Fleeson, 2001; Fleeson & Jayawickreme, 2015). In an experience sampling (ESM) study, Fleeson (2001) assessed the distribution of Big-Five personality states in daily life and found that although average behavioral tendencies are stable over time, people consistently manifest behaviors across the spectrum of the personality dimensions. For example, someone who is introverted will typically engage in behaviors along all levels of the introversion-extraversion spectrum in their daily routine, presumably in response to situational demands or goals.

Experience sampling studies have also observed that extraverted states correspond with higher levels of positive affect, irrespective of trait personality (Wilt et al., 2012). The phenomenon of traits and states being characterized by similar outcomes (e.g., state and trait extraversion correlated with positive affect) is referred to as the state-trait isomorphism hypothesis (Fleeson, 2001; van Allen & Zelenski, 2018). This isomorphic effect has also been observed in laboratory settings where researchers instruct people to behave in either an extraverted (e.g. acting bold, assertive, adventurous, energetic) or introverted manner (e.g. acting reserved, quiet, passive, unadventurous). When interacting in a group discussion, people who acted extraverted reported significantly higher positive affect compared to those who acted introverted, regardless of their trait personality (Fleeson et al., 2002; Zelenski et al., 2012). In other words, acting extraverted is associated with increases in positive affect, regardless of whether people are dispositional introverts or extraverts (Fleeson et al., 2002; McNiel et al., 2010; McNiel & Fleeson, 2006; Zelenski et al., 2012; Zelenski, Whelan, et al., 2013).

Despite concerns that acting in a manner inconsistent with one’s disposition may tax one’s mental resources and be perceived as being inauthentic, research has shown that extraverted behavior is associated with elevated levels of subjective authenticity. Of particular note is that this effect holds irrespective of disposition (Fleeson & Wilt, 2010). Moreover, acting extraverted does not typically produce negative cognitive or affective consequences, nor is it perceived as more effortful, for introverts (Gallagher et al., 2010; Zelenski et al., 2012). Interestingly, introverts nonetheless underestimate the affective benefits, and overestimate the affective costs, of acting extraverted (Zelenski, Whelan, et al., 2013).

Based on these findings, researchers have suggested that acting extraverted may be an effective means of increasing the well-being of dispositional introverts (Fleeson et al., 2002; McNiel et al., 2010). However, to date, the strategy of enacted extraversion has largely been confined to short-term, controlled laboratory settings, and has focused on a narrow set of outcome variables (e.g., affect, cognitive fatigue). Outside the lab, experience sampling studies suggest that introverts experience more positive affect when behaving in extraverted ways, but results are correlational and ambiguous about whether additional, intentional or instructed extraverted behaviors would have the same results as lab studies. To date, only a very limited set of (lab) contexts have been studied experimentally. Beyond the basic issue of generalizability, it is plausible that the clear instructions and pre-defined interaction partners in lab studies make it easier and more pleasant for introverts to behave in counter-dispositional ways. That is, pairing introverts with conversational partners in a contrived setting removes a natural barrier for introverts, namely, seeking out and establishing social contact. Our goal was to bridge the gap between the lab and the real world, testing whether instructions to act extraverted would increase positive affect, and possibly other aspects of well-being, in day-to-day life. The answer is an important step in determining the viability of an extraversion-based happiness exercise.

As we prepared this paper, we became aware of two recently published articles reporting on similar experiments. In one recent study Margolis and Lyubomirsky (2020) instructed participants to engage in extraverted behavior for one week followed by one week of introverted behavior (and vice versa; i.e., a within-person manipulation). The behavior manipulation adjectives (i.e., talkative, assertive, spontaneous vs. deliberate, quiet, reserved) were selected with the goal of minimizing social desirability. Using structural equation models and second-order latent growth models in a sample of undergraduate students, results supported previous observational and experimental data in finding that extraverted behavior increased positive affect and that this effect was not moderated by trait levels of extraversion. Additionally, extraverted behavior also led to increases in connectedness and autonomy (as assessed via the balanced measure of psychological needs; Sheldon & Hilpert, 2012), and flow (Csikszentmihalyi, 1990). Weaker effects were observed on negative affect, subjective happiness, and life satisfaction.

In another similar study (Jacques-Hamilton et al., 2019), participants (n=147) were randomly assigned to an enacted extraversion or ‘sham’ condition. A sample (n=76) from an unrelated study was also used as a no-contact control group. Importantly, in this experiment participants were instructed to act more “bold, talkative, outgoing, active and assertive” (extraverted condition) or “unassuming, sensitive, calm, modest, and quiet” (sham condition) in their ’interactions with other people across the next week”. Participants reported on behaviors, moods and authenticity six times daily, at the end of the week, and two weeks later. Overall, the instructions to act extraverted produced more positive affect and feelings of authenticity than the comparison condition in daily and end-of-week reports. The amount of time spent in social interactions did not mediate the relationship between positive affect and extraverted behavior, though all participants were asked to alter their behavior in the context of social interactions. Surprisingly, this study also suggested—for the first time—that benefits might depend on dispositional extraversion. The positive affect boost was not present for extremely introverted participants, who also reported some negative affect, tiredness, and inauthenticity when asked to act in more extraverted ways (Jacques-Hamilton et al., 2019). This new finding suggests some caution in recommending extraverted behavior. It also underscores the importance of testing whether patterns change when going from correlational and lab studies to real world interventions, and these novel results require confirmation or refutation via replication.

The present research extends the methodology of previous enacted extraversion studies to the context of people’s everyday lives via three, week-long behavioral interventions. Specifically, our three studies assess whether engaging in an extra 15 minutes of extraverted behavior each day for five consecutive days is sufficient for increasing positive affect in the moment and producing lasting effects over time. In Studies 1 and 2, we compare the outcomes associated with engaging in daily extraverted behaviors to engaging in daily introverted behaviors (Study 1) and a control task (Study 2). In Study 3, we further extend the enacted extraversion paradigm to test whether engaging in social extraverted behavior produces different outcomes than non-social extraverted behavior or an inert control activity. Our studies were designed and conducted before Jacques-Hamilton et al. (2019) and Margolis et al. (2020), yet still differ in some useful ways: we had participants form implementation intentions (Gollwitzer, 1999), plan specific, concrete, short-term activities for each day (vs. general acting instructions for a week); we include different comparison (control) conditions; and we investigated extraverted behavior beyond socializing (especially Study 3). Based on the established finding that enacted extraversion leads to positive affect in laboratory settings, we propose the following hypotheses:

Extraverted behavior will result in greater feelings of positive affect ‘in the moment’ (i.e., via responses reported during daily reports), relative to introverted behavior and control groups.

Extraverted behavior will lead to increases in positive affect from the pre-test assessment to the post-test assessment.

Despite the recent finding suggesting an interaction between trait extraversion and extraverted behavior in predicting positive affect (Jacques-Hamilton et al., 2019), many other null results led to our a priori hypothesis that dispositional levels of trait extraversion would not influence the effect of extraverted behavior on positive affect. Still, we acknowledge that our studies were not designed (powered) well to conclusively find or refute plausibly sized interactions as they were conducted prior to the Jacques-Hamilton et al. (2019) report.

As the majority of research in this area has focused on the relationship between positive affect and enacted extraversion, we focus our hypotheses on this association. However, we also present exploratory analysis examining potential outcomes of an ‘acting extraverted intervention’. For example, we explore whether variables that have been studied in the enacted extraversion paradigm, such as authenticity (Whelan, 2014) and perceived effort (Gallagher et al., 2010) are influenced by the instructions. Additionally, we explore indicators of well-being that are traditionally not incorporated within enacted trait research—but that are more common among happiness exercises—such as meaning in life, subjective happiness, life satisfaction, depression, and vitality. Finally, we include a measure of serenity to assess the empirically dubious, but commonly asserted, suggestion that introversion might produce low-arousal pleasure (cf. Zelenski, Sobocko, et al., 2013). As recommended by Simmons and colleagues (2012), we report how we determined our sample sizes, all data exclusions, all manipulations, and all measures in the studies herein.

Seventy-three undergraduate students participated in the study in exchange for course credit and entry into a cash draw. Our initial sample size target (n=100) was based on a ‘rule of thumb’ of 50 participants per condition. We did not stop/add participants based on results available during data collection; recruitment was limited by the end of the academic term. Data from 66 participants who completed a minimum of three out of five daily activities were included in the analysis. Thirty-five participants were randomly assigned to the enacted extraversion condition and 31 were randomly assigned to the enacted introversion condition. A CONSORT diagram (Schulz et al., 2010) for study 1 is available in Figure 1. Participants were primarily female (71%), with ages ranging from 18 to 45 years old (M = 20.69, SD = 6.45). Sensitivity analysis, conducted in G*Power (Erdfelder et al., 1996), suggest that at 80% power our sample could detect effects sizes of d = .31 and d = .62 for dependent and independent samples t-tests, respectively and f = .19 $(ƞ2=.03)$ for a between-within ANOVA interaction. Lab studies have found large differences in positive affect between conditions with instructions to act extraverted vs. introverted (e.g., d = 1.30, Zelenski et al., 2012). We acknowledge that there is substantially less power to detect interactions (i.e., with dispositional extraversion), limiting our ability to conclude the hypothesized nulls.

Participants registered for a study entitled “Examining the relationship between well-being and behavior”, which was advertised as consisting of an in-lab session (pretest), five daily activities and corresponding online surveys (daily logs 1-5), and a follow-up online survey (post-test) in exchange for course credit and the opportunity to be entered into a cash draw for completing activity reports.

Participants were randomly assigned to one of two conditions (acting extraverted or acting introverted). During the in-lab pre-test, participants completed self-report questionnaires to assess personality, affect (including positive affect, negative affect, serenity, and fatigue), authenticity and effort, meaning in life, subjective happiness, satisfaction with life, depression symptoms, and subjective vitality. Following completion of the questionnaires, participants completed a written task that presented them with five behavior-related adjectives (either extraverted adjectives or introverted adjectives). Participants listed ten ways in which they could embody these characteristics in the context of their daily lives. Following verification of the list’s content by the researcher, participants then chose five activities to enact over the following five days and assigned each activity to a specific day of the week. In the five days following the lab-session, participants were sent identical daily surveys. These surveys inquired about the extent to which participants engaged in their enacted activity. Only those who indicated that they either completed, or somewhat completed, their daily activity were asked to respond to self-report assessments of their mood, vitality, and feelings of authenticity and effort while engaging in their assignment. Those who indicated that they did not complete their activity responded to an identical question set, but in the context of how they felt over the past day. These reports were excluded from analysis.

One week following the initial in-lab session (i.e., one day following the final daily activity), participants completed an online post-test questionnaire to assess their overall subjective well-being and affect.

Personality. To assess personality during the pre-test questionnaire, we administered the Big Five Inventory (BFI), consisting of 44 items, that is designed to assess the five higher-order personality dimensions within the Five Factor Model: extraversion, agreeableness, openness, conscientiousness, and neuroticism (Costa & McCrae, 1992; Digman, 1990; Goldberg, 1990; John et al., 1991). Participants rated how much they agreed with statements such as “I see myself as someone who is talkative” or “I see myself as someone who can be reserved” as being characteristic of them on a scale of (1) Disagree strongly to (5) Agree strongly. Each personality dimension was scored by calculating the mean response across the items in each scale¹.

Well-being Measures. To assess well-being during both the pre-test and post-test questionnaires, participants responded to several measures including the Positive and Negative Affect Schedule (PANAS), the Meaning in Life Questionnaire (MILQ), the Subjective Happiness Scale (SHS), the Satisfaction with Life Scale (SWLS), the Depression Scale (CESD), the Subjective Vitality Scale (SVS), and the Authenticity and Effort Scales (AES).

Positive and Negative Affect Schedule (PANAS). We used a 44-item version of the PANAS-X questionnaire to assess participants’ average weekly mood (Watson & Clark, 1994). Participants described the extent to which the mood adjectives described their feelings over the course of the previous week on a scale of (1) I felt that way very slightly or not at all to (5) I extremely felt that way. The full PANAS-X questionnaire contains 60 items. To suit the needs of our study, we omitted items from the ‘basic negative emotion scales’ and included items from the ‘general dimensions scales’ (measuring positive and negative affect), the ‘basic positive emotions scales’ (measuring joviality, self-assurance, and attentiveness), and the ‘other affective states’ scale (assessing shyness, fatigue, serenity, and surprise). However, only positive affect (PA), negative affect (NA), fatigue, and serenity were used in the present analysis. Example items include: “active”, ”alert”,”enthusiastic” (PA); “ashamed”,”nervous”, “distressed” (NA);”tired”, “sluggish” (fatigue); and “relaxed”, “at ease” (serenity).

Meaning in Life Questionnaire (MILQ). The MILQ consists of ten items, with two subscales to assess both one’s presence of and search for meaning in life (Steger et al., 2006). The 5-item presence subscale assesses the extent to which the respondent feels that they currently have a sense of meaning in their life, while the 5-item search subscale assesses the extent to which the respondent is actively seeking to obtain a sense of meaning in their life. Participants rated items including “I understand my life’s meaning” (presence) and “I am looking for something that makes my life feel meaningful” (search) from (1) Absolutely false to (7) Absolutely true. The MILQ was scored by calculating the mean response across each item pertinent to its respective subscale.

Subjective Happiness Scale (SHS). The SHS consists of four short items to measure subjective happiness (Lyubomirsky & Lepper, 1999). For this measure, participants responded to questions on a scale of 1-7 to describe the statement which they felt best described them. An example included participants rating to what extent they were (1) Not a very happy person or 7) A very happy person. The SHS was scored by calculating the mean response across all four items.

Satisfaction with Life Scale (SWLS). The SWLS consists of five items that assess global life satisfaction (Diener et al., 1985). Participants rated the extent to which they agreed with statements on a 1-7 scale, including “In most ways my life is close to ideal” and “The conditions of my life are excellent”, whereby a rating of 1 indicated that they strongly disagreed with the statement and a rating of 7 indicated that they strongly agreed. The SWLS was scored by computing the sum of all five items.

Subjective Vitality Scale (SVS). The SVS consists of six items that measure subjective vitality, defined as the feeling of being alive and alert (Ryan & Frederick, 1997). Participants rated six statements based on how they applied to their life over the course of the past week, on a scale of (1) Not at all true to (7) Very true. The SVS was scored by calculating the mean of the six items.

Authenticity and Effort (AE). The AE questionnaire consists of two scales that assess subjective feelings of authenticity and effort over the course of the past week (Fleeson et al., 2002; Whelan, 2014). A 15-item version of the scale was administered, instructing participants to indicate how they felt over the course of the past week. On a scale of 1-7, participants rated items including “It was very easy to behave the way I did” (effort) and “I was my true self” (authenticity), whereby a rating of 1 indicated that they strongly disagreed and a rating of 7 indicated that they strongly agreed. Authenticity and effort were scored by calculating the mean for each scale’s respective items.

The Center for Epidemiologic Studies Depression Scale (CESD). The CESD (Radloff, 1977) is a short self-report measure designed to measure depressive symptoms in the general population. The scale presents 20 first person statements such as “I felt depressed”, “I felt lonely” and “I thought my life had been a failure”. For each item the respondent is asked to identify how often they have felt in a manner consistent with the statement presented over the period of the past week. The response options were rarely or none of the time (less than one day), some or a little of the time (one to two days), occasionally (three to four days) and most of the time (five to seven days). Scores on the CESD range from 0 (no depressive symptoms) to 60; a score of 16 or below is considered to indicate “no clinical significance”. Following convention, the CESD score was calculated by using the sum of all 20 responses.

Behavioral Instructions. During the pre-test, participants completed a short writing task to prepare for the activities that they would engage in during the following week. In the extraversion condition, participants were given a list of five trait descriptive adjectives coinciding with extraverted behaviors (assertive, energetic, outgoing, sociable, and adventurous), and in the introversion condition, participants were given introverted adjectives (reserved, cool/aloof, quiet, modest, and passive). Trait adjectives for both conditions were derived from Goldberg’s (1990) work on the big five and represent a breadth of facets for each construct. However, trait adjectives for the introversion condition were intentionally chosen for their positive connotations in order to minimize differences in social desirability between conditions. For example, aloof was chosen over unsociable, reserved over secretive, passive over submissive, and unflattering terms such as unadventurous, lethargic and joyless were avoided all together. This selection bias was employed to address the criticism that the language used to describe introversion and extraversion are themselves biased (Zelenski, Sobocko, et al., 2013).

Participants in both conditions were then instructed to “list 10 ways in which you could embody one or more of these adjectives for a 15- minute period in the context of your upcoming week.”). Examples of extraverted activities listed by participants (after completion) included, “I went out to have some food with my girlfriend, and I was being the one who talks”, “was extra friendly with my co-workers”, and “I was to ask my prof about volunteer opportunities after class; we chatted for 5 minutes”. Examples of introverted activities included “I practices yoga for about 15 minutes”, “went out with friends, but I left really early”, and “at work…let others pick what we would do.”

After the list of potential activities was verified by the researcher, participants were given a blank five-day calendar and asked to assign one approved activity to each day. Additional instructions were given to indicate which time of day they intended to engage in the chosen activity. This aspect of the intervention was informed by research on the effectiveness of implementation intentions. Implementation intentions are thought to facilitate the goal intention-behavior relationship, as they provide a process of how goal intentions are translated in behavior (Gollwitzer, 1993).

Activity Description. Each day, participants began by reporting the extent to which they completed their daily activity (“did complete”, ”somewhat completed”, or”did not complete”). Participants who completed or somewhat completed their activity proceeded to provide a brief description of the activity that they engaged in that day, including its time and duration. These participants were then shown the well-being measures and activity ratings and were asked to apply these measures to how they felt during their activity that day.

Participants who did not complete their activity were not prompted to describe their activity, but instead responded to the remainder of the questionnaire.

Well-being Measures. During the daily questionnaires, participants responded to three measures of well-being that were also included in the pre- and post-test questionnaires (vitality, PANAS, authenticity/effort), but adapted to refer to how they were feeling during their respective daily activity. To prevent daily fatigue within participants when completing each daily questionnaire, we used a revised 27-item version of the PANAS (containing only items assessing PA, NA, fatigue, and serenity) rather than the 44-item version used within the pre- and post-test questionnaires (Watson et al., 1988), and a 6-item adapted version of the Authenticity and Effort questionnaire rather than the 15-item version used in the pre- and post-test questionnaires. All measures were scored using the average ratings for items in a given scale.

To test whether enacted extraversion was associated with elevated, ‘in the moment’, levels of positive affect (PA), relative to introverted behavior, we conducted an independent samples t-test (see Table 1 for means and standard deviations for all variables). Consistent with previous findings, participants in the enacted extraversion condition reported higher levels of daily positive affect than participants in the enacted introversion condition, t(64) = 4.79, p = .001, d = 1.18. Thus, Hypothesis 1 was supported.

To determine whether the intervention lead to changes in positive affect over time, A 2 x 2 mixed ANOVA on positive affect was conducted with condition (extraversion, introversion) as the between-subjects factor, and time (pretest, post-test) as the within-subjects factor. ANOVA results for all variables measured at pre- and post-test can be found in Table 2. Results revealed a significant time by condition interaction, F(1, 54) = 8.56, p .01 , partial ƞ² = .14. Follow-up tests showed that positive affect increased in the extraversion condition, t(28) = -2.31, p .05, 95% CI [-.82, -.04], but not in the introversion condition, t(26) = 1.84, p = .08, 95% CI [-.04, .75]. Neither the main effect of time, F(1, 54) = .09, p = .76, nor the main effect of condition, F(1, 54) = 2.46, p = .12, were statistically significant. Thus, Hypothesis 2 was supported in this study.

To determine whether trait extraversion moderated positive affect experienced during extraverted behavior, we ran a sequential regression analysis with dispositional extraversion and condition predicting positive affect (step 1) and a trait extraversion by condition interaction (step 2). In the full model, condition significantly predicted PA (β = -0.85, p .001), but trait extraversion (β = -0.04, p = .80) and the condition by trait extraversion interaction (β = .11, p = .64) did not. Thus, results failed to reject our null Hypothesis 3; i.e., they were consistent with expectations, but with the caution that nonsignificant results do not provide strong evidence for a null conclusion.

Exploratory analyses (refer to Tables 1-3) yielded several interesting findings. For example, participants in the extraversion condition reported feeling more authentic during their condition specific activities than did those in the introversion condition. These findings are consistent with experience sampling data (Fleeson & Wilt, 2010) and other experimental manipulations in a real-world setting (Jacques-Hamilton et al., 2019). Additionally, those who enacted extraverted behavior reported elevated levels of vitality relative to their counterparts in the introversion condition on daily reports and when comparing pre-post assessments. To our knowledge, this is the first reported instance of experimentally induced extraverted states producing increases in subjective vitality; however, that the association between trait extraversion and vitality (Ryan & Frederick, 1997) holds at the state level is consistent with the trait-state isomorphism hypothesis (Fleeson, 2001). Lastly, there were indications that extraverted behavior decreased symptoms of depression and fatigue over time compared to introverted behavior, but other well-being comparisons did not yield significant differences (e.g., weekly changes in life satisfaction or subjective happiness). Finally, we did not observe any positive effects of introverted behavior in comparison to extraverted behavior (e.g., even feelings of serenity).

Study 1 sought to extend lab-based findings to participants’ daily lives and test the efficacy of a five-day enacted extraversion intervention as a strategy for increasing positive affect. Consistent with our hypotheses and previous experimental and observational findings, participants reported more positive affect after engaging in extraverted behaviors than introverted behaviors, and this effect was not dependent upon dispositional levels of extraversion (cf., Jacques-Hamilton et al., 2019). Further, and consistent with our second hypothesis, participants in the extraversion condition (vs. introversion condition) reported higher levels of positive affect after the week of the intervention compared to the before the intervention.

Study 1 was limited in part by a small sample size and the absence of a neutral control condition. To address these concerns, we conducted a second experiment (Study 2) with modifications to the experimental design and recruitment strategy. Specifically, we sought to deliver a similar intervention, with a non-introversion control condition, delivered solely online (cf., the in-person delivery of implementation intentions in Study 1) to boost sample size.

Two hundred and seventeen undergraduate students participated in Study 2 in exchange for course credit and entry into a cash draw. Our initial target sample size was 200 (i.e., 50 per condition, plus expected attrition), however we continued to collect data until the end of the academic term. Nine participants failed at least six of eight attention checks and were removed from all analysis. Two hundred and eight participants completed the pre-test and at least three daily logs and were included in analysis of daily outcomes (randomly assigned to extraverted behavior [n = 71], introverted behavior [n = 73], or a control [n = 64]). One hundred and seventy-eight participants completed both the pretest and posttest, as well as at least three daily logs, and were included in pre-post analysis (ns = 58, 65, 55 for extraverted, introverted, and control conditions, respectively). A CONSORT diagram (Schulz et al., 2010) for Study 2 is available in Figure 2. The full sample used in daily assessments (n = 208) was primarily female (72%), with ages ranging from 17 to 46 years old (M = 21.96; SD = 5.34). Sensitivity analysis suggest that in mixed ANOVAs our sample could detect, with 80% power, effect sizes of f = .22 $(ƞ2=.05)$ for between subject tests, f = .08 $(ƞ2=.01),$ for within subject tests, f = .09 $(ƞ2=.01)$ for between-within interactions, and f = .23 $(ƞ2=.05)$ for a one-way ANOVA.

In Study 2, participants began by completing an online pre-test questionnaire similar to Study 1 but delivered completely online (versus in lab). Participants were randomly assigned to either the acting extraverted, acting introverted, or control condition. Participants in both the extraverted and introverted conditions were presented with five behavior related adjectives, descriptive of either extraversion or introversion (mirroring Study 1) and were asked to write down five different ways which they could embody these characteristics for fifteen-minute periods each day for five consecutive days. Participants were instructed to choose activities or behaviors that they had not already planned to be part of their week. In the control condition, participants were asked to list five early childhood memories that they could write about each day for the following five days.

For the five days following the initial questionnaire, participants completed a brief daily questionnaire which assessed mood, vitality and authenticity during their assigned activity. In the experimental conditions, participants were asked to provide a brief summary of their daily assignment, while the control condition, participants were asked to write about an early childhood memory for approximately fifteen minutes.

One day following the final daily survey a follow-up (post-test) questionnaire was administered. This questionnaire contained the same measures as the initial questionnaire, and as the post-test questionnaire in Study 1. An additional set of questionnaires was administered two-weeks following the post-test assessment; due to high attrition rates these data are not reported here.

For Study 2, the materials used for the pre- and post-test questionnaires, and daily questionnaires, mirrored those used in Study 1, with the addition of modified behavioral instructions (adapted to the online context and adding control group) during the pre-test questionnaire.

During the pre-test, participants were randomly assigned to either one of the two experimental conditions similar to Study 1 (extraverted or introverted), or to a new control condition introduced within Study 2. Those in the experimental conditions completed a short writing task to prepare for the activities that they would engage in during the following week, mirroring the task in Study 1. In the new control condition, participants were instructed to write about an early childhood memory for fifteen minutes each day (Seligman et al., 2005).

A one-way ANOVA on daily positive affect scores indicated differences between conditions, F(2, 204) = 6.25, p = .002. Follow-up tests showed that average positive affect during the condition-specific activities was significantly higher in the extraversion (M = 3.12; SD = .70) condition than the introversion (M = 2.70; SD = 7.1), t(141) = -3.57, p = .001, 95% CI [-.65, -.19], but not the control (M = 2.98; SD = .77), t(132) = -1.14, p = .26, 95% CI [-.39, .11]. Participants in the introversion condition also reported less positive affect than did those in the control condition, t(135) = 2.19, p = .03, 95% CI [.03, .53]. Thus, Hypothesis 1 received only mixed support.

A 2 (time) x 3 (condition) mixed ANOVA was conducted on self-reported positive affect from pre- to post- assessment. Descriptive statistics and ANOVA results are presented in Tables 5 and 6, respectively. Results revealed a significant main effect of time, F(1, 175) = 11.40, p .001, and a time by condition interaction F(2, 175) = 5.05, p .01 (see Table 6). A series of repeated measures t-tests were employed to follow up on the observed interaction. Follow-up tests revealed that positive affect scores decreased in the introversion condition, t(64) = 4.43, p = .001, 95% CI [.20, .52]. No statistically significant differences were evident over time in the extraversion, t(57) = -.04, p = .97, 95% CI [-.16, .15], and control condition, t(54) = 1.46, p = .15, 95% CI [-.05, .31]. Thus, Hypothesis 2 received only mixed, at best, support.

Similar to Study 1, to determine whether trait extraversion influenced PA experienced during enacted behaviors, we ran a sequential regression analysis with dispositional extraversion and condition (extraversion and introversion vs. control) predicting PA (step 1) and then adding trait extraversion by condition interactions (step 2). Results from the full model revealed that trait extraversion (β = 0.22, p = .04), and the introversion condition (β = -.28, p = .02) predicted positive affect significantly; however the extraversion condition (extraversion: β = .14, p = .26) and key condition by trait interactions (extraversion condition β = .12, p = .44; introversion condition β = -.16, p = .32) were not statistically significant. Thus, results were generally consistent with our null Hypothesis 3, but with the caution that nonsignificant results do not provide strong evidence for a null conclusion.

Exploratory analyses suggested that, similar to Study 1, momentary extraverted behavior was associated with higher levels of subjective authenticity and vitality compared to introverted behavior (see Tables 4-6). However, extraverted behavior did not produce more vitality than in the control condition, and authenticity was highest in the control condition. Further, we observed that authenticity increased over time in the control condition, decreased over time in the introversion condition, and did not change significantly in the extraversion condition. Presence of meaning in life increased for both the extraversion and control conditions but not in the introversion condition, However, acting extraverted did not have a significant impact on many other well-being indicators (e.g., life satisfaction, subjective happiness, negative affect, etc.) over the course of the intervention.

In Study 2, we sought to replicate our first experiment while strengthening our approach with the addition of a control condition and an increased sample size. In line with our first hypothesis, we found that participants reported more daily positive affect following extraverted behavior than introverted behavior. However, there was not a statistically significant difference between positive affect levels reported in the extraverted behavior and control condition (i.e., reflecting on an early childhood memory). Moreover, the weekly assessment of positive affect was not significantly higher after a week of acting extraverted exercises, though positive affect in the introversion condition decreased to a statistically significant degree. Collectively, the results of Study 2 are more limited in suggesting well-being benefits of acting extraverted in day-to-day life when compared to Study 1. Some of the apparent benefits seem to depend on the particular comparison condition. Jacques-Hamilton et al. (2019) found somewhat similar patterns (i.e., differences depending on the particular control comparison).

The particular control task (i.e., writing about early childhood memories) is one of many possible reference groups against which the effects of enacted extraversion and introversion could be compared. Interestingly, participants in the control condition reported increases in the presence of meaning in life over the course of the intervention, and they also reported elevated levels of authenticity ‘in the moment’. Although the early childhood memories writing task has been used in past research as an inert control (Seligman et al., 2005), real changes in authenticity, affect, and meaning following this nostalgia also seem plausible (e.g., Stephan et al., 2012), as do demand explanations. For these reasons, we opted to adjust the control condition in Study 3.

In Study 3, we sought to replicate and extend our enacted extraversion intervention design by testing whether the effect of extraverted behavior on positive affect would vary as a function of whether the behavior was social or non-social in nature. The propensity for extraverts to engage in more social interactions has long been hypothesized to account for at least some of the discrepancy between positive affect levels experienced by introverts and extraverts (e.g., Eysenck & Eysenck, 1985; Pavot et al., 1990). However, this perspective has also been challenged (e.g., Lucas et al., 2008; Margolis & Lyubomirsky, 2020). Indeed, this hypothesis was tested via a mediation model by Jacques-Hamilton et al. (2019) who found no association. We sought to experimentally test whether positive affect was differentially associated with positive affect for social vs. non-social extraverted behavior. We therefore omitted the introversion condition in this study to focus on the varieties of enacted extraverted behavior more closely, and included a new, less meaningful control condition: reporting on daily events.

Two hundred and sixty-five undergraduate students participated in Study 3 in exchange for course credit and entry into a cash draw. Mirroring Study 2, we sought to obtain 200 participants, with the goal of obtaining 50 participants per condition after accounting for expected attrition; however, we continued to collect data until the end of the academic term. Eleven participants did not complete the pretest assessment, thirty-nine participants failed at least six of eight attention checks and were removed from all analysis. Two hundred and fifteen participants completed the pre-test and at least three daily logs and were included in analysis of daily outcomes (randomly assigned to social extraverted behavior [n = 68], non-social extraverted behavior [n = 69], or a control [n = 78]). A CONSORT diagram (Schulz et al., 2010) for study 2 is available in Figure 3.

One hundred and fifty-seven participants completed both the pretest and posttest, as well as at least three daily logs, and were included in pre-post analysis (ns = 50, 52, 55 for social, non-social, and control conditions, respectively). The full sample used in daily assessments (n = 215) was primarily female (73%), with ages ranging from 17 to 53 years old (M = 22.11; SD = 6.66). Sensitivity analysis suggest that in mixed ANOVAs our sample could detect, with 80% power, effect sizes of f = .23 $(ƞ2=.05)$ for between groups, f = .10 $(ƞ2=.01)$ within subject tests, f = .11 $(ƞ2=.01)$ for between-within interactions, and f = .25 $(ƞ2=.06)$ for a one-way ANOVA.

Similar to Study 2, participants began by completing an online pre-test questionnaire with measures to assess baseline personality and well-being, before being randomly assigned to one of three conditions. More specifically, participants were assigned to either the social extraverted, non-social (independent) extraverted, or control condition. Participants in the experimental conditions were asked to plan five days of activities according to the trait descriptive adjectives provided (i.e. either social or non-social adjectives associated with extraversion), and either in the context of others (social), or independent of others (non-social). Participants in the control condition were instructed to reflect on the time period between 12 and 12:15 pm each day for the following week. This control condition differed from Study 2 to better mirror the activity reflection sessions within the experimental conditions, but with no additional behavioral intervention.

For the five days following the pre-test questionnaire, participants completed daily online questionnaires similar to those used in Studies 1 and 2. On the final day of the experiment (one week after the pre-test), participants completed a post-test questionnaire with questions mirroring the pre-test session. An identical questionnaire was administered two weeks following the post-test assessment; however, data from this response set is not presented due to low response rates.

All questionnaires from the two previous studies were included in Study 3, with the addition of four scales to assess constructs relating to sociability and novelty. Modifications were also made to the behavioral instructions to reflect the updated nature of the daily activities and control condition.

In addition to the measures used in Studies 1 and 2, participants completed the Social Connectedness Scale - Revised (SCS-R; R. M. Lee et al., 2001) and the Inclusion of Other in the Self Scale (IOS; Aron et al., 1992). These scales were not included in the present analysis.

In addition to the daily measures used in Studies 1 and 2, participants completed daily ratings assessing novelty (adapted from T.-H. Lee & Crompton, 1992) and sociability (adapted from Keyes, 1998). To assess daily novelty, participants responded to four items developed by the researchers based on adjectives representing novel behaviors, adapted from the four dimensions of novelty: thrill, alleviation of boredom, change from routine, and surprise (T.-H. Lee & Crompton, 1992). Examples included “I experienced something that I never have before”, and “I did something that was out of the ordinary for me”. Participants responded to these questions based on how they felt during their daily activity on a scale of (1) Strongly disagree to (7) Strongly agree. Items were averaged into a ‘novelty’ score. To assess daily levels of sociability, participants responded to five items adapted from Keyes’ (1998) Social Well-Being Scale (SWBS), which were used by Smillie et al. (2015) in the context of extraversion interventions. The SWBS is divided into five subscales, four of which were used within the context of the present study: social integration, social acceptance, social contribution, and social actualization. During the daily questionnaires, participants responded to four questions corresponding with each of the subscales, as well as one additional question to assess social context specifically (“There were many people around me”). Examples of items from the subscales included “I had meaningful interactions with others” (social actualization) and “I felt close to the other people around me” (social integration), with participants rating each question on a scale of (1) Strongly disagree to (7) Strongly agree. The five items were averaged for a ‘social’ score.

During the Study 3 initial session, participants who were randomly assigned to one of the two experimental conditions completed a short writing task to prepare for the activities that they would engage in during the following week. In the social condition, participants were given a list of five characteristics associated with extraversion involving social interaction (talkative/outgoing, sociable, assertive, playful, dominant), and were asked to “Plan five activities that they would engage in over the following five days within the presence of other people” (i.e., within a social context). In contrast, participants in the non-social condition were given a list of five adjectives related to characteristics of extraversion that did not involve social interaction (energetic/lively, adventurous, spontaneous, confident, bold), and were asked to”Plan five activities to engage in during the following five days without the presence of other people” (i.e., independently). Participants in both conditions were encouraged to choose activities that were outside of their everyday routine.

We changed the control condition in Study 3 (cf. Study 2) to instruct participants to reflect and report on what they were doing between 12-12:15 pm each day for the next five days. This reflection acted as a comparable daily ritual that mirrored the experimental conditions’ reflection on daily activities, but without introducing new daily behaviors into their day-to-day lives.

To ensure that participants in the three separate conditions were engaging in the desired activity, manipulation checks confirmed that those in the social extraverted condition were indeed behaving significantly more social (daily ratings) than those in the control condition, t(115) = 5.88, p 0.01, d = 1.10 and those in the non-social condition, t(117) = 5.44, p .001, d = 1.01 (see Table 7 for descriptive statistics). No significant differences existed between social ratings for those in the non-social condition and the control condition. Moreover, participants in the non-social condition engaged in activities that were significantly more novel compared to those in the control condition, t(122) = 10.76, p 0.001, d = 1.93, and those in the social condition, t(117) = 3.11, p 0.01, d = 0.57. Surprisingly, participants in the social condition also reported engaging in more novel behaviors compared to those in the control condition, t(115) = 7.65, p 0.001, d = 1.42.

A one-way ANOVA of positive affect reported during daily tasks suggested significant differences among the experimental conditions, F(2, 212) = 7.53, p .001. Follow-up tests showed that participants reported less positive affect in the control condition compared to both the social, t(144) = -3.39, p .001, 95% CI [.-65, -.17], and non-social, t(145) = -3.17, p .01, 95% CI [-.63, -.14] conditions. Social and non-social groups reported similar levels of positive affect, t(135) = -.21, p = .83, 95% CI [-.26, .21], supporting previous research suggesting that acting extraverted produces increased positive affect and Hypothesis 1.

A 2 (time) x 3 (condition) mixed ANOVA revealed a main effect of time on positive affect, F(1, 155) = 13.79, p .001, partial ƞ² = .08, with positive affect decreasing on average during the study period. A significant interaction effect was also observed, F(2, 155) = 3.21, p = .04, partial ƞ² = .04. Follow-up t-tests showed that positive affect scores decreased over time in the control, t(55) = 3.56, p = .001, and non-social conditions, t(51) = 2.12, p =.04, but did not significantly change in the social condition, t(49) = .55, p = .59. The results did not support Hypothesis 2, but the relatively more positive trajectory for social extraversion might suggest some incremental benefit to these behaviors.

A regression analysis similar to the model used in Study 2 was used to test the impact of dispositional extraversion on PA experienced ‘in the moment’ in each of the three conditions. Results from the full model revealed that while the social (β = .40, p .001) and non-social (β = .35, p .001) extraversion conditions predicted PA, trait extraversion (β = .06, p = .65), and interactions between trait extraversion and both the social (β = .29, p = .09) and non-social (β = .21, p = .23) conditions were not statistically significant predictors. Dispositional extraversion did not significantly moderate the effect of acting extraverted (social and non-social) on positive affect. Results were thus generally consistent with Hypothesis 3, but with the caution that nonsignificant results do not provide strong evidence for a null conclusion.

Exploratory analyses suggested that extraverted behavior was associated with elevated levels of daily subjective vitality compared to the control condition. Subjective vitality experienced during the daily tasks was significantly lower in the control condition compared to both the social and non-social extraverted conditions. Subjective vitality did not significantly differ between extraverted conditions. Interestingly, participants did not report any significant differences in daily subjective authenticity or feelings of serenity between conditions. When examining changes from pre- to post-test well-being, there were several effects of time, but there were minimal significant differences when considering the more interesting interactions between time and condition.

In Study 3, we found that participants in both extraverted conditions (social and non-social) experienced higher positive affect than in the control condition. Moreover, there were no significant differences between the different extraverted conditions on level of daily positive affect. These results suggest that the hedonic benefits of behaving extraverted do not necessarily depend on interpersonal factors (Cooper et al., 1992; R. M. Lee et al., 2008). In possible contrast to this conclusion, positive affect did decrease less over time (pre to post) in the social extraversion condition, compared to both the non-social and control conditions. It is possible to see this as a beneficial buffering effect of social extraversion. That is, while daily non-social extraverted behaviors seem effective in increasing daily positive affect, daily social behaviors may have a longer lasting effect that persists over time. On the other hand, given the generally null pattern of condition by time effects on well-being, our manipulation seems to impact daily outcomes more so than weekly, longer-term, outcomes. Positive affect decreasing over time in all conditions also seems to argue against the effectiveness of acting extraverted exercises on weekly positive affect. As such, we put somewhat more stock in the daily results which suggest similar short-term affective benefits for both social and non-social extraverted behavior. Finally, consistent with our original hypothesis, and the results of Studies 1 and 2, dispositional levels of trait extraversion did not significantly moderate the extent to which those in either the social or non-social extraversion conditions reported elevated levels of positive affect.

Although we focused our analysis on primary hypotheses relating to positive affect, here we briefly summarize intervention effects across our three studies on both positive affect and additional outcome measures of well-being. For these exploratory analyses, we use statistical significance to flag potentially interesting findings; however, our studies were not powered to test so many hypotheses, and we do not correct for multiple comparisons. Appropriate caution is warranted. Table 10 summarizes results for between-group comparisons of how participants felt ‘in the moment’, and within-subject effects from pre-test to post-test assessments. Notably, momentary effects show greater positive affect and subjective vitality in extraversion conditions compared to both introversion (Studies 1 and 2) and control (Study 3) conditions and show greater feelings of authenticity when participants acted extraverted rather than introverted (Studies 1 and 2). Effects observed over pre- and post-test assessments were generally weak and inconsistent. Except for Study 1, enacted extraversion did not increase positive affect over time. Across conditions and studies, negative affect decreased over time. Counterintuitively, depressive symptoms tended to increase over time while participants also tended to report more presence of meaning in their lives after completing the studies.

To explore possible trait by condition interactions for all measured indicators of well-being, we conducted a series of regression analyses, with the standardized beta coefficients (trait, condition, and interactions) reported in Table 11. For a more descriptive approach to exploring possible interactions, we also examined patterns in correlations between trait extraversion and each outcome variable, separately by condition (Table 12). The correlation comparisons across conditions can hint at interactive patterns (or lack thereof) similar to the way a table of means by conditions suggests similarities and differences beyond inferential tests.

In Studies 1 and 2, trait extraversion was negatively correlated with perceived behavioral effort for those who engaged in extraverted activities. Said another way, introverts were more likely to perceive extraverted behavior as requiring more effort, suggesting a potential cost to acting counter-dispositionally. Further, all studies found a positive correlation between authenticity and trait extraversion in the acting extraverted condition, and smaller or reversed correlations in the comparison conditions. This again suggests that introverts may experience costs when acting extraverted. However, correlations are inconsistent across studies, and the cross-product terms which test the patterns in regression are rarely significant.

Overall, despite correlational findings suggesting that acting counter-dispositionally may incur some costs (i.e., feeling inauthentic, requiring greater effort), tests of interactions in regression models did not provide strong support in either direction. With limited power to detect small interaction effects, the null results are not high confidence.

This research further explores the robust link between extraversion and positive affect with a more novel focus on intentionally adding extraverted behaviors to daily life. Consistent with observational data, laboratory-based experiments, and real-world interventions, across three studies, we observed that engaging in extraverted behavior was associated with elevated levels of in-the-moment positive affect compared to introverted behavior and/or other control behaviors (Hypothesis 1). However, general, retrospective reports suggested that participants did not consistently experience increased positive affect from extraverted behavior during the week of the intervention, compared to the preceding week (Hypothesis 2). Further, we did not find clear support for the idea that dispositional extraverts might benefit more from engaging in more extraverted behavior compared to dispositional introverts (Hypothesis 3). These three studies suggest that average levels of momentary positive affect can be increased through the addition of extraverted behavior; however, these results do not support the idea that enacted extraversion interventions of 15-minutes daily behavior change are sufficient to produce lasting changes in well-being.

One strength of our studies was the randomized controlled trial designs in Studies 2 and 3. Interestingly, the effect of extraverted behavior on positive affect in our studies varied depending on the control group. We observed the strongest effect of extraverted behavior when compared to enacted introversion (Studies 1 and 2) or journaling about daily happenings (Study 3). In contrast, when compared to an arguably more ‘active’ control, reflecting on one’s childhood (Study 2), the relative benefits of extraverted behavior were not clear. Taken together with conceptually similar experiments, increases in positive affect appear to be most consistent when compared to enacted introversion (present studies; Margolis & Lyubomirsky, 2020) or tasks which contain several elements of introverted behavior (i.e., a ‘sham’ condition which included instructions to act unassuming, sensitive, calm, modest, and quiet; Jacques-Hamilton et al., 2019). Said another way, some of the suggested benefits of extraverted behavior could be equally well-described as costs of introverted behavior.

When considered with the available observational (e.g., Fleeson et al., 2002; McNiel et al., 2010; McNiel & Fleeson, 2006) and experimental data collected both in laboratory (e.g., Zelenski et al., 2012; Zelenski, Whelan, et al., 2013) and naturalistic settings (e.g., Jacques-Hamilton et al., 2019; Margolis & Lyubomirsky, 2020), our results contribute to the confidence in a robust effect of state extraversion on positive affect. Although we did not find strong evidence that this effect varied as a function of trait extraversion (cf., Jacques-Hamilton et al., 2019), correlational evidence in the present studies did suggest that trait introverts perceived extraverted behavior to require more effort, and that trait extraversion was (inconsistently) related to feelings of authenticity when completing extraverted tasks. These interactions were generally non-significant in regression models, though even our larger studies may be underpowered to detect interactions reliably. To be clear, we did not design our studies to test null trait by condition interactions given the pattern of null interactions in the existing literature. Following planned analysis, we considered the application of equivalence tests (e.g., Lakens et al., 2018; Schuirmann, 1987) to determine if we could rule out interactions that were larger than trivially small (i.e., similar to concluding the null). Recently, the equivalence testing framework has been expanded to apply to interaction terms in multiple regression (Alter & Counsell, 2021). However, post-hoc applications of equivalence testing are problematic, and the size of our samples were not well suited to the use of equivalence testing for multiple regression (Alter & Counsell, 2021). Detecting significant interactions, and persuasively ruling them out, both require substantial power. Additionally, equivalence testing requires the identification of a SESOI (smallest effect size of interest) and after careful consideration we were unable to arrive at a reasonable SESOI based on available research given the scarcity of similar studies employing multiple regression and difficulty of translating ‘just noticeable differences’ to interactions including a continuous (trait) predictor. Ultimately, we believe that any reasonable SESOI would be likely to produce an ‘inconclusive’ result due to our sample sizes. Therefore, while we cannot rule out the possibility that the enacted extraversion-positive affect association depends on dispositional extraversion, we did not observe strong evidence for its influence. Our research is ultimately inconclusive regarding interactions, and future work will likely need substantially larger samples (and to define a SESOI a priori) to provide more definitive answers.

It is also possible that we did not observe stronger evidence of ‘costs’ to acting counter-dispositionally (cf. Jacques-Hamilton et al., 2019) due to methodological differences between studies. For example, participants in our experiments were asked to engage in fifteen extra minutes of introverted or extraverted behavior each day (mirroring the length of time used in lab-based studies), whereas Jacques-Hamilton and colleagues (2019) asked participants to modify their behavior in interactions with others “as much as possible” for one week, and Margolis et al. (2020) likewise instructed participants to ‘be as introverted/extraverted as you can’. Although comparisons of the duration and intensity of extraverted behavior are not possible between studies, it is likely that people engaged in extraverted behavior less frequently in our studies.

Notably, fewer trait interactions have been observed when participants have been asked to report on specific instances of behavior, for example via experience sampling methods or questionnaires on one’s day. Specifically, the moderating effect of trait extraversion on positive affect in the Jacques-Hamilton et al. (2019) study was observed in retrospective accounts but not in momentary data. The difference in task instruction within the present studies (i.e., 15 minutes) may make it more reasonable for introverts to introduce short bursts of extraverted behavior into their daily lives, rather than trying to adjust their general approach when interacting with other people (which would naturally be easier for dispositional extraverts). Therefore, dispositional levels of extraversion may influence the costs and/or benefits derived from extraverted behavior, at least for positive affect, when such behavior is enacted more frequently and is sustained over longer durations of time. Further research is required to determine the trade-offs between short-term and long-term counter-dispositional behavior and the hedonic benefits of sustained enacted extraversion.

Collectively, the accumulating evidence from observational and experimental data supports the trait-state isomorphism hypothesis (Fleeson, 2001). The trait-state isomorphism hypothesis also predicted the association between subjective vitality and state extraversion observed in each of our three experiments. Specifically, the association between trait extraversion and vitality (Ryan & Frederick, 1997) is predicted by the isomorphism hypothesis to hold at the state level, an association we observed consistently. However, given the strong correlations between measures of positive affect and subjective vitality, both measuring high-arousal and high-pleasantness constructs, the tendency for these outcomes to co-vary with state extraversion is perhaps not surprising and has been observed previously (Pickett et al., 2020).

Although consistently observed through a variety of methods, the mechanisms underlying the state extraversion/positive affect association are still not well-understood. One potential hypothesis for this association is that the relationship between extraversion and social interactions may partially explain this link. We attempted to dissociate social from non-social forms of extraversion in Study 3 with separate instructions. Participants who added socially motivated extraverted behavior to their daily lives (e.g., “I made plans with a friend and consciously made the effort to be more talkative and outgoing…”) did not report any more positive affect than did participants who engaged in non-social extraverted behavior (e.g., ”After finishing work, I took the bus down to the market and tried Ethiopian food, I went alone and made my way around the market until I found something that was new to me”). This finding is consistent with mediation analysis (Jacques-Hamilton et al., 2019) which did not detect a mediating role of social activity frequency and the positive affect/state extraversion relationship. Taken together, these results suggest a possible ‘alternative path’ for dispositional introverts who wish to elevate their positive affect through primarily non-social activities or for situations in which social contact is not feasible.

Our three experiments provide experimental confirmation that intentionally incorporating more extraverted behavior into one’s daily life can increase momentary levels of positive affect. However, four main limitations should be noted. First, the studies are each limited by low compliance rates, with many participants excluded from analysis for failing to complete a minimum number of daily activities or for non-compliance. This could be a result of the length of the study and/or loss of interest over the course of the week (Lefever et al., 2007). However, we cannot rule out the possibility that some degree of selective participation (i.e., person-activity fit) was present and may potentially bias the results, or limit generalizability in ways that would decrease the efficacy of a broad intervention. Additionally, although we attempted to assess participants two weeks following the post-test with the goal of assessing longer-term changes in well-being, significantly high attrition prevented analysis of these data.

Second, the experimental designs in Studies 2 and 3 were implemented entirely online, which we suspect resulted in lower task compliance and the selection of activities which may not be ideal for the purposes of the studies. In our first study, participants’ activities were discussed and selected with the assistance of a researcher; however, the transition to an entirely online intervention (which facilitated scalability) resulted in a reduced capacity for quality control. Although it is tempting to attribute the stronger results observed in Study 1 to this characteristic of the experimental design, especially when identifying significant pre-post changes in well-being, the smaller sample size and possible experimenter effects cannot be overlooked as being limiting factors.

Third, the ‘dosage’ of the experimental manipulation may have been insufficient to produce detectable changes in affect and well-being over time. One tentatively supporting piece of evidence for this suggestion comes from a recent 15-week study where participants who engaged in more enacted extraversion tasks were more likely to report positive changes in trait extraversion over time (Hudson et al., 2019). Although this experiment focused on trait change rather that well-being outcomes, it highlights the possibility that a higher frequency of enacted extraversion may be required for sustained changes in trait-related outcomes (e.g., positive affect, well-being).

Fourth, although the promise of enacted extraversion as a well-being increasing strategy is to provide individuals with a simple behavioural tool to increase their mood and well-being, our experiments are conducted at the group level and results cannot be inferred at the individual level. The findings from our two randomized controlled trials provide some evidence that average levels of positive affect may be enhanced at the group level. However, without further researching using idiographic methods, we must refrain from committing an ecological fallacy (Robinson, 1950) by suggesting that our experiments support enacted extraversion as a strategy for any given individual. Moreover, across the recent acting extraverted studies, there appears a possible trade-off between the amount of activity needed for lasting change and the emergence of concurrent costs for dispositional introverts (cf., Jacques-Hamilton et al., 2019).

Finally, although participants in all conditions received similar behavioural instructions, it is possible the behavioural adjectives of extraversion were more socially desirable than those for introversion and influenced participants’ expectations regarding the purpose of the study. Behavioural instructions in the introversion conditions were designed to counter this potential effect, however, it is unclear whether the study outcomes were influenced by social desirability.

Multiple sources of data (e.g., observational, lab-based, randomized-controlled-trials) converge on the phenomenon that extraverted states produce elevated levels of positive affect as experienced in the moment. This effect was independent of dispositional levels of extraversion in three intervention studies and does not appear to be determined by the sociability of extraverted behavior (Study 3), suggesting an alternative ‘non-social’ route for introverts to increase positive affect. Further, in two of three experiments, extraverted behavior was reported as feeling more authentic than introverted behavior or control tasks. However, the addition of only fifteen minutes of extraverted behavior did not produce enduring effects on positive affect or other indicators of well-being. We suspect that more frequent and sustained extraverted behavior may be required to elicit enduring effects, though more research is needed to confirm whether sustained counter-dispositional behavior can also be as beneficial for well-being as the momentary acts. Future research should address these questions, as well as work on implementation details, to capitalize on the thus far encouraging potential for an acting extraverted happiness exercise.

Contributed to conception and design: ZvA (S1/2), DW (S3), TS (S1), JZ (S1-3)

Contributed to acquisition of data: ZvA (S1/2), DW (S3), TS (S1), JZ (S1-3)

Contributed to analysis and interpretation of data: ZvA (S1/2), DW (S3), TS (S1), JZ (S1-3)

Drafted and/or revised the article: ZvA, DW, JZ

Approved the submitted version for publication: ZvA, DW, TS, JZ

None

All the stimuli, presentation materials, participant data, and analysis scripts can be found on this paper’s project page on the Open Science Framework: https://osf.io/czauv/